PR Summary

This PR introduces a new demonstration Jupyter notebook and a suite of tests that expand the ValidMind library’s analytical capabilities. The demo notebook (plots_and_stats_demo.ipynb) provides end-to-end examples for exploring data using both visualization and statistical analysis methods. It showcases interactive plots such as correlation heatmaps, histograms, box plots, and violin plots, alongside statistical tests including descriptive statistics, correlation analysis, normality tests, and outlier detection.

New test modules have been added under two categories: "plots" and "stats". The plotting tests include modules for rendering box plots, correlation heatmaps, histograms, and violin plots. The statistical tests extend functionalities to compute descriptive statistics, perform correlation analysis with significance testing, run multiple normality tests, and detect outliers using several methods (IQR, Z-score, and Isolation Forest).

Each module introduces thorough parameter validation, comprehensive documentation within docstrings, and clear error handling (e.g., using SkipTestError for invalid input scenarios). The test provider registry has been updated to include the new categories ensuring that the tests integrate seamlessly with the existing ValidMind test framework. Overall, this update enhances exploratory data analysis capabilities and provides a robust, interactive demonstration for end users and developers alike.

Test Suggestions

• Run the demo notebook end-to-end to verify that all plots and statistical outputs render successfully and interactively.
• Unit test each new module with valid data to ensure that all customization options (e.g., colors, grouping, threshold parameters) are applied correctly.
• Validate error handling by providing datasets with missing required columns or constant values to trigger SkipTestError exceptions.
• Test edge cases such as datasets with very high dimensionality or non-numeric columns to ensure graceful degradation.
• Verify that the registration of test providers now correctly includes the new 'plots' and 'stats' categories.

Adding @juanmleng and @johnwalz97 to collect some feedback. I like the idea of reusable plots and stats but I wonder if they should be tests or maybe we should think about another primitive that can be reused by any test? 🤔